SpecMRL: Redesign Matryoshka Representation Learning within Parametric Spectrum Paper • 2610.26824 • Published 24 days ago • 33
What Really Works for Universal Multimodal Embeddings? A Controlled Study of Contrastive Training Strategies Paper • 2610.15035 • Published Oct 10 • 14
LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation Paper • 2610.19975 • Published 29 days ago • 43
Learning to Read Out: Unembedding Dynamics in Language Model Pretraining Paper • 2610.21025 • Published 29 days ago • 6
LEV: Thinking Fast and Deep with Looped Decision Models Paper • 2610.20951 • Published 29 days ago • 72
Periodic Weak Spots: Phase Sensitivity from Chunked KV-Cache Compression Paper • 2609.36322 • Published Sep 28 • 36
Redesign Mixture-of-Experts Routers with Manifold Power Iteration Paper • 2606.12397 • Published Jun 10 • 50
Your UnEmbedding Matrix is Secretly a Feature Lens for Text Embeddings Paper • 2606.07502 • Published Jun 5 • 55
Coupling Experts and Routers in Mixture-of-Experts via an Auxiliary Loss Paper • 2512.23447 • Published Dec 29, 2025 • 67